development and use of a mathematical … · viour in collision configurations where the experiment...
TRANSCRIPT
DEVELOPMENT AND USE OF A MATHEMAT I CAL MODEL S IMULATING
A TRAFFIC ACCI DENT V I CT IM
D . Lestrel i n , A . Fayon , C . Tarri ere , Laborato ire de Phys i o l ogi e et de B iomecani que PSA/Renaul t
La Garenne-Col ombes ( France ) .
One of the a i ms of th is paper i s to b i gh l i ght a p h i l osophy for emp l oy i ng mathematica l model s . The other a im i s to present some resu l ts obtai ned i n the fi e l d of uti l i zat ion of the �VMA-APR mathemati ca l model drawn from three studi es on parti cu l ar probl ems shown in the appendi ces :
- two studi es on �edestri an confi gurat i on , - and one i n head-on col l i s ion confi guration wi th a pas senger restrai ned
by a shoul der be l t and a � bar .
HSRI D ES I GNED I N 1973 A TWO-DIMENS I ON , ei ght-mass mathemat ica l model , ca l l ed " MVMA-2D" . The Peugeot-Renau l t As soci at ion 1.-;as ab l e to acqu i re i t , and deve l oped i t for the i r account . The programme was i n i ti a l l y devel oped and used i n the " occupant" confi gurati on ( head-on col l i s i on ) , and a l so i n the " pedestri an " con fi guration . F i na l l y , the model was extended to the " s i de col l i s i on " confi guration .
One shou l d not be surpri sed that certai n i mprovements made to the model s i nce i t was des i gned converge , i f a compari son i s made between the work conducted at the Uni vers i ty of Mi chi gan ( 1 ) ( 2 ) ( 3 ) , and our own resu l ts . The APR modi fi cati ons were pri ority ori ented by the need for a mode l i zat ion of human behavi our in col l i s i on confi gurations where the experiment was cri ti cal and the theory 1 i tt1 e known .
·
I - THE F I ELD OF APPL I CAT ION OF MATHEMATI CAL MODELS ( M . M)
Four types of app l i cations of mathemati ca l model s can be d i sti ngui shed , each requi ri n� spec i a l precautions a nd a n i ncreas ing degree of val i dati on , i n the fol l ow ing order :
a ) He lt i n understandi ng the phenomena - The user , a fter defi n i ng the s imul ated col 1 s 1 on data himself, has at füs d i s posal a set of resul ts whi ch i s a perfectly l og i ca l sequence of a cause and effect chai n . Even though the res u l t may not be perfectly rea l i s t i c a s certai n aspects of rea l i ty are roughly s i mul ated o r negl ected , everyth ing that hap�ens can b e exp l a i ned and t h i s exp l anation i s acces s i bl e , e i ther i n the mode l concepti on , or i n the s imu lat ion resu l ts or data . I n parti cul a r , a l l accel erations of the v i cti m ' s body segments can be exp l a i ned by the forces generated i n contacts , by the centri fuga l effect of rotat ion , by the art i c u l at ion coupl es , atc . A l l these coup l es and forces can i n turn be exp l a i ned by the pos i ti on and vel oc i ty of the dummy wi th re l at ion to the veh i cl e at any moment i n time . A mathematica l mo-
47
del can therefore be an excel l ent teach i ng i nstrument , provi ded that i t functi ons we l l , that the schemati zati on i t implies does not di stort real i ty too muc h , that the res ul ts a re fai rly comp l ete ( forces , vel oci t i es , accel erati on crump l i ng , coup l es , etc . ) and i s stated i n a c l ear and "reada b l e " manne r . The user shoul d a l s o make the effort to fi nd the exp l anati ons to a l l the questi ons that may present themsel ves .
b ) Ca l cul ati n$ phys i ca l l y unmeas urab l e parameters - When conduct ing experi mental tests , it i s easy to measure accelerat1 ons. Movements , forces and couples are more d i ffi c ul t ( at l east for some ) . For each extra meas urement , there i s a correspond ing extra cost ( cost of sensor or measuri ng i nstrument , fi xture , record i n g , di smantl i ng , ana lys i s , ca l cul ati ons , etc . ) As for some q uanti t i e s s uch as k i neti c energy , amount of movement , crus h i ng energy , they are i mpos s i b l e to measure . On the other hand , as the resul ts s upp l i ed by a mathemati cal model are pure products of ca l cul ati o n , a l l var i a b l e functi ons a re acces s i b l e or can be cal cul ate d . The user has far too much to choose from , and the o utputs a re general ly contro l l ed by a seri es of se l ector swi tches . Shoul d the user rea l l y need a parti cul ar functi on whi ch i s mi s s i ng from the desi gned l i s t of outputs , i t i s a l ways poss i b l e for the ca l cu l a t i on and the pri ntout of thi s functi on ' s va l ues to be added to the model ( Cf . l i s t of vari a b l es added as further output) . A mathemati ca l model ( M . M . ) can therefore serve to comp l ete meas urements made i n experimental testi ng : the mathemati ca l model data shoul d therefore be adj us ted so as to make the res ul ts of the s i mul ati on coi n c i de wi th the known and meas ured resu l t s of experi mental testi ng ( trajecto ri es , acce l e rati ons , etc . ) As soon a s the resu l ts coi nc i de s at i s factori l y , a ny parti cul ar functi on can be ca l cul ated ( see examp l e i n Appendi x l: ca l cu l ati ng energy di stri buti on
-pas s i ve bel t ) . Such a use of M . M . can a l s o be envi saged for acci dent recons tructi ons . The conditi ons for u s i n g a M . M . i n th i s manner are as fo l l ows : - a suffi c i ent n umber of resul ts shoul d be had on the reference test or the
acci dent to be reconstructed ( trajectGr i es , i mpact poi nts , l evel s of accel erati on , etc . ) so that reproduci ng the known resu l ts wi l l ens ure that the unknown resul ts and data are reproduce d . Examp l e : know i ng the coordi nates of the i mpact poi nt of a pedestri an ' s head on a veh i c l e wi l l not s uffi ce to correctly s i mul ate the acci dent i f a l l the other parameters are not known ( s i ze of pedestri an , veh i c l e vel oci ty at moment of i mpact , etc . ) On the other han d , knowi ng the s i ze of the pedestrian can enabl e the impact ve l oci ty to be eval uated us i n g s i mu l ati ons i n wh i ch the veh i c l e vel oci ty wi l l be gradua l ly i ncreased or decreased unti l the ri ght i mpact l ocat i on i s fo und .
I n other words , the more re l i ab l e data the user has at h i s di sposa l , the l es s ri s k there i s of the effects of certa i n errors o n the i nput data be i ng cancel l ed and consequently be i ng h i dden from him because of the effects of other errors . - The model shou ld ens ure s uffi ci ent q ua l i ty of s i mu l at ion so that the user
may reproduce q u i te accurate l y the res u l ts of a test . - F ina l l y , the user s houl d take pai ns and ca l l on h i s i ntui t ion i n order to
defi ne the most re l evant set of i n put data . H i s task wi l l be made eas ier i f he has sets of data approachi ng the confi gurati on he wi s hes to s i mu l ate .
c ) Study of the i nfl uence of a parameter - Th i s i s what M . M . are us ua l l y used
48
for . The techni q ue consi sts , fi rst l � , i n fi r a 1 i : i ng a set of reference data enabl i ng a s imul at ion to be produce ..: 1·1: · 1 i s co�si dered rea l i s t i c and re-presentati ve of a co l l i s i on the user L 1 ,·: -c�re::i ted i n , and then i n chang ing the val ues of certai n i nput parameters � . . � i n fl uence of wh i ch i s to be stud ied , wi th the others keepi ng thei r reference val ues . The res u l t s are then compared wi th the reference resu l ts , thus enab l i ng the i nfl uence of the i nput data , the va l ue of wh i ch has changed , to be quanti fi ed taki ng the resu l ts as a bas i s ( maximum rates of accel erat ion , severi ty i ndi ces , trajectory amp l i tude , forces sustai ned , etc . ) See appended examp l es : 1 1Study of the i nfl uence of bumper pos i t i on on the k i nemati cs of a pedestri an 1 s head when pedestri an i s struck 11 , 11Study of the i nfl uence of severa l dummy parameters on i ts propens i ty to submari ne 11 • The condi t ion for a mathemat i cal model to be used i n such a manner is that the model be c l ose enough to rea l i ty for us to rely on the sense and amount of output resul t vari ati ons . The model and set of reference data shoul d therefore be va l i dated for a t l east one confi gurat ion c l ose to that be i n g s i mu l ated .
d) Oi rectly predi cti ng test resu l ts - , even i f there are no experimental data i n s 1 m1lar condit1ons . Th1 s 1 s the most ambi tious man ner of u s i ng a M . M . l t cons i s ts i n predi ct i ng , wi th j ust one s imu l at ion , the trajectories and i mpact ve l o c i t i e s wi th accuracy , and the l evel s of acce l eration and severi ty i ndi ces l ess accurately i n the case of col l i s i on , based on a l i mi ted amount of synthet i c i nformation ( anthropometri c data of v i c ti m , i mpact vel oc i ty , s hape o f veh i c l e dece l eration curve i n a head-on col l i s i on , approxi mate pos i ti o n of vi cti m , etc . ) We a re present ly work i ng i n th i s di rect ion on s i mu l ati ng pedestri a n col l i s i ons , and the resu l ts obta i ned enab l e us to be reasonably confi den t . The condi t i o n for such a use i s : - to spl i t up the i nput parameters i nto two categori es : those for wh i ch the
a verage and common val ues can be frozen for a l l the tests ; those for wh i ch the val ue shou l d be accurately determi ned for each parti cu l a r confi gurati on .
- Ass i gn standard va l ues to the parameters of the fi rst category . - Val i date the model wi th the thus-defi ned data over a re l ati ve ly h i gh num-
ber of experimental col l i s ions as di fferent as poss i b l e .
I I - REFLECT IONS ON THE NOTION OF VAL I DAT I NG
Val i dat ion i s never acqui red absol ute l y . At a certai n s tage i n devel opment of the model ; we have a s tate of va l i dat ion wh i ch i s l i mi ted to : - certai n confi gurat ions ( e . g . a model can be va l i dated for the occupa n t and
badly val i dated for a pedestri an , or we l l va l i dated for 3-poi n t be l ts a nd badly val i dated for di fferent pas s i ve bel t s ) ;
- certa i n fi e l ds of var i at ion of i nput parameters ( e . g . for pedestri a n : l arge vehi c l es , l ow veh i c l es , sma l l and ta l l pedes tri ans , l ow , a verage a nd h i gh speeds , parti cu l ar s hape of veh i c l e front ) .
- Certai n output parameters ( e . g . a good q ua l i ty of trajectory s imu l ati on wi l l be obtai ned more eas i ly than the accel eration l evel s on i mpact ) .
These restri ctions wi l l depend on the degree of accuracy requ i red by the user on the parameters he i s i nterested i n . They wi l l a l s o depend on the data for feed i ng the model as much as on the mode l i tsel f : i f the va l ues of certa i n i nput data can be readi ly determi ned by measurement , others are di ffi cu l t to be determi ned experimental l y .
Thus , a model i s progressi vely val i dated as new col l i s i ons a re s imu lated , certai n parts of the model a re i mproved upon , and experi ence we have on data
49
makes progres s .
I I I - WORK CARRI ED OUT ON THE MODEL
I I I - 1 . - Qua l i ty of s i mu l ati on . -
For a n equi val ent uti l i zati on cost , the advantages o f a model i ncrease wi th the qua l i ty of s i mul at ion i t affords . Thi s just i f ies efforts be i ng made on the model when i t appears that s uch and s uch a modi fi cation , whether i t be p i n-poi nt or fundamenta l , wi l l enab l e a cons i derabl e i mprovement of the model behavi our to be hoped for i n a preci se s i tuati on or i n al l cases . Thi s i s a l ways reasonabl e when the i ntended modi fication i s based on a more thorough ana lys i s of the phenomena . On thi s score , at l e ast three parts of the mode l have been i mproved upon i n parti cul ar .
Bel t system i n head-on col l i s i on - Step by s tep , we have deve l oped a be l t system w1th opt1 ons enabl1 ng different confi 0urati ons to be s i mul ated real i s t i ca l l y . These options a re : - 3-po i nt bel t wi th buckl e and s l i di ng be't ts , or i ndependent l ap and shoul der
be l ts , or both types of 2-poi nt be l ts ; - a dherence or s l i p w i th rubb ing at thorax , buck l e and pel v i s l eve l s , - s imul ati on of the be l t between top anchor po i nt and reel , wi th or wi thout
rubbi ng at return l oo p , - s i mul ation of be l t t i ghten i ng i n reel , - uni form d i s tr i buti on of bel t stretch i ng over who l e l e ngth of bel t , or l ocal
deformati on i n a l oad l i m iter , - pos s i bi l i ty of anti c i pated ten s i on i ng of bel t by a ti me-control l ed retrac
tor. - poss i bi l i ty of cancel l i ng bel t-thorax t i es in order to keep good be l t geo
metry . Art i cu l at i ons - The fri cti on coup l e model and the stop coup l e mode l have
been modified provi di ng more rea l i s ti c 1 1dummy 11 behavi our and better stabi l i ty of the step-by-step i ntegrat i ons of programme . Muscu l ar reacti ons coup le s have been added as opti on to s i mu l ate the v i ct ims of real acci dents .
Angl e probl ems i n veh i c l e contact surfaces - Thi s del i cate probl em of generating real1 st1 c contact forces when 1 mpact1 ng s harp edges , wh i ch i s i mpor- · tant for pedestri ans , has been deal t wi th by d i s t i ngu i s h i ng bet\veen two types of contact a reas wi th d i fferent " behav i ours " : "bent s heet meta l " types of reg ion ( bumper , front nose panel , bonne t , w i ndscreen , etc . ) and reg i ons the s hape of whi ch l i mi ts a ful l vol ume ( paddi ng , soft-nose , etc . )
Thi s model i zat i on g i ves sati s facti on , a l though i t can s ti l l be i mproved upo n .
I I I - 2 . - Improvi ng usabi l i ty . -
When the model becomes eas i er to use and when the res u l ts become more 11mean i ngfu l 1 1 , the users w i l l u se the model more readi l y , l ose l ess t ime by us i ng i t , and draw more i nfo rmat ion from i t .
Th i s justi fi es the efforts made on the fo l l owi ng deta i l e d mai n poi nts : Re l i ab i l i ty of operati on - Executi on end ing abnorma l l y for reasons other
than data errors have become very rare . "Safety devi ces " have been added . Even i f there i s a probl em , the res ul ts al ready obtai ned are pri nted out and p l otted due to the programme bei ng d i v i ded i nto i ndependent s tages ( s imul ati on ,
50
ca l cul at ion of i n di ces , pri nti ng out of rcs u l t� and p l otti ng of curves , s i mul at i on d i s p l ay ) .
S imp l i fyi ng data - Certa i n data wh i ch are di ffi c u l t to determi ne have been rep l aced by automatic ca l cu l ati ons .
Further pri ntout outputs - Outputs have been added to those al ready exi s t i ng . They are destined elther to hel p i n understandi ng the s imul at ion sequence , or i n compl et ing s i mul ati on resul ts .
Cal cu l at i ng severi ty i nd i ces - Cal cul at ing and pri nti ng out i ndi ces i n an i n dependent stage from simulating has been conducted with great care . Bes i des ca l cul at i ng the sever i ty i nd i ces on components A-P and S- 1 , and on the resu l tan ts of head and thorax accel erati ons , cal cu l at i ng the t ime duri ng whi ch the s pec i fi ed acce l e rat ion thresho l ds of 0 to 200 g are exceeded i n steps of 1 g , for head , thorax and pe l v i s accel erati ons , can be made . The resul t i ng l evel s of acce l erat i on exceeded for 3 ms are cal cul ated for the head , thorax a nd pel vi s . A very performi ng H I C cal cu l ati on has been des i gned . I ts executi on t i me does not i ncrease with the square of the number of poi nts , but approximately wi th the square root of the number . Furthermore , i t a utoma t i ca l l y cal cul ates the H I C correspond ing to each suffi ci ently i so l ated and pronounced peak ( e . g . pedestri an i mpacts wi th veh i cl e and ground ) .
Di spl ay i ng k i nemati cs - The v i ct i m i s drawn by a p l atter i n h i s envi ronment (vehicle profile , occupant be l t i f any , the ground for pedestri a n ) , at the sca l e requ i red by user , i n a reference frame re l ated to the veh i c l e or to the ground wi th the sequences bei ng separated or superimposed ( see fi gure 1 ) . The p l ot can be made a utomat i ca l l y at the same t ime as the s i mul ati on i mmedi ate ly after the resu l ts have been pri nted out . The user can , furthermore , obta i n data on cards at o utput whi ch can be stored i n the a rchi ves , thus enabl i ng h im to have other di spl ays i n o ther condi t i ons ( other scal es , offboard datum i ns tead of separate , etc . ) To compare the trajectori es o f two tests , the users can make d i s p l ays on a transparent support and s uperimpose them or have them p l otted on the same sheet wi th d i fferent l i nes . These d i s p l ays are extremel y useful i n understandi ng the phenomena ari s i ng duri ng a s i mu l ated col l i s i o n , and i n i nterpreti ng the resu l ts .
Di sp l ay and check of i n i ti a l cond i t i ons pri or to ca l cul ati ng - The set of data can be tested prior to carry1 ng out the s 1 mulat 1on using a programme that traces the dummy , the veh i c l e and , i f requ i red , the bel t , i n the i ni ti a l col l i s i on pos i t i on at 1/5 sca l e , and can a l so show up certa i n anoma l i es i n the set of data .
Pl otti ng of output curves - Al l the resu l ts i t i s pos s i bl e to pri nt out at o utput can be select1 vely plotted on a vari a b l e n umber of sheets . On each sheet, the user can have a l l the var i a b l e he wi shes traced out as ordi nates depend i ng on the same parameter as the absci ssa , whi ch may be t i me , but a l so any other parameter , taken from a same category or d i fferent categori es , i n order to compare the shape of the curves , h i ghl i ght any dephas i ng or synchroni sms , determi ne graph i ca l l y equi va l ent masses , trace trajectori es of e l l i pse/veh i cl e contact poi nts , effort/deformat i on curves , etc .
I V - I NTERACTIONS BETWEEN DEVELOP I NG AND EXPLO IT ING THE MODEL
l t goes without sayi ng that worki ng on the programme wi l l i mprove the qual i ty of the resul ts obtai ned by users . For examp l e , the i ncreas i ng s i mi l ari ty whi ch i s obta i ned i n occupant confi gurati o n i n a head-on col l i s i on between the mathemati cal s imul ati ons and experi mental reference tests can be exp l a i ned on the one hand by greater data experi ence , and on the other by succes s i ve impro-
5 1
VEHJ LtJLE - .:l(JGiJJT 20 4 . P A vl T'St: :- 4_ KM/H
.#IN. �E 1...U: N = .�lt: T JN JflR�JLlJ A.JiE
':1:' . .
OE L I LS.�Al :: 2/u uENJU.\ P'JLS :.ir:u�JiT.R = E�HEU.�
..iA TE = 041 J1173 �QilLE = t140
t = 0 r= D·O ll5iC
1
T' dl·O ll!lC
T= �k ll!l�
.!;""'�
.. . .
Fi g . 1
7
. = �t).O 16 .c
8
;. a. • . i&:r. f: ..fu·O ISf.C
9
'� t = 100 ms
:: 1.i).Q .l!H T- Ull O &e
10
\� .t t = 200 ms
„ 1·0.0 .&D: � z.o.o s ·c
Exampl e of ki nemati c d i s p l ay - separated sequences ( above ) - superimposed sequences ( oppos i te )
5 2
vements made to the programme , notab ly t o the bel t model . But i t shou l d be s tressed tha t , conversely , not on ly i s the explo i tat ion of a l l the model s vi tal i n order to h i gh l i ght operat ing probl ems a1 ;d test programme modi fi cations - the tests , a l though necessary , wi l l not suffice - but i t i s a l so very useful for showi ng up a uti l i ty h ierarchy among al l the i mprovements i t wou l d be tempti ng to programme . I t enab l es therefore the person i n charge of i mprovi ng the programme to carry out the most useful work i n pri ori ty .
CONCLUS I ON
The resul ts devel oped i n the appendi ces wi l l not be recal l ed ; i t i s rather a quest ion of taki ng stock of the use made of MVMA/APR .
After several years of practi ce wi th a rel at i vel y compl ex mathemati cal model , i t appeared that resul ts wh i ch coul d be veri fied on l y wi th d i ffi cu l ty i n experi ments , coul d be obta i ned us i ng the M . M . Does t h i s mean that the proport i on of research work conducted us i n g model s wi l l i ncrease i ndefi ni tely to the detriment of research work conducted experimenta l l y ? No answer can be g i ven to thi s quest i on for the choi ce of ei ther approach wi l l depend on the permanent adaptat i on work whi ch i s necessary both for the mode l s and ful l -s i ze acci dent s imul ations . The mode l s need s i mul at ion qual i ty , experi ments need repeti ti vi ty . I t can however be remarked that any extend i n g of the domai n access i bl e to the model means conducti ng val i dat i on tests and that the test resul ts have repercuss i ons on the model i nput data and pose questi ons rel at i n g to the l atter.
REFERENCES
1- D . H . ROBB INS , R . O . BENNETT , and B . M . BOWMAN , " MVMA Two-Dimen s i onal Crash V i ctim S i mul ati on , Vers i o n l . 11 Fina l Report to Motor Veh i c l e Manufacturers ' Assoc i ati o n , UM-HSRl -B l -73- 5 .
2- B . M . BOWMAN , R . O . BENNETT , and D . H . ROBBI NS "MVMA Two-D imensi onal Crash V i ct i m S i mul ati on , Vers i o n 3� Uni vers i ty of M ich i gan Report No . UM-HSR I B l - 74- 2 , Ju ly 1974 .
3- D . H . ROBB INS , B . M . BOWMAN and R . O . BENNETT : 11The MVMA Two-D imensi onal Crash V i ctim S i mul ation " , i n Proceedi n gs of the 18th Stapp Car Crash Conference , p p . 657-678, New-York , S . A . E . I nc . , 1974 .
53
A P P E N D I X I
STUDY OF THE I NFLUENCE OF BUMPER POS IT ION ON THE KINEMATI CS OF A
STRUCK-PEDESTRIAN ' S HEAD . (Apri 1 1977 )
Quest ion : Wi l l a bumper , desi gned to reduce the ri s k of s eri ous knee i nj u ry to a s truck pedestri an , due t o i ts pos i t i on l ower than the kneecap , i n crease the ri s k of fatal head i nj ury because of the i ncrease i n head/veh i cu l e i mpact vel oci ty or of a l es s favourabl e l ocat ion of the i mpact poi nt ?
1 - Simul at ion data .
Pedestri an : 50 percenti l e , upri ght Impact s peed : 32 km/hr ( 20 mph) Veh1 cule : two profi l es were used :
a l ow streaml i ned profi l e ( that of the C i troen GS ) a " square " profi l e ( that of the Opel Kadett model 66 , type B )
Bumper : The s ame for both veh i cl es ( s hape , s trength )
On ly the hei ght ( h ) and overhang ( d ) wi th re l at i on to front end of the veh i c l e vary and are defi ned i n fi g . 2 . I n thi s pape r , the pos i ti on d = O i s ca l l ed "norma l " , and d 'f 0 i s cal l ed "projecti on" .
2 - Di scus s i on on data . 2 . 1 - The i mpact vel oc i ty chosen i s that at whi ch fatal pedestri an acci
dents appear . 2 . 2 - Pos i t i on of veh i c l e body i s that obtai ned wi th the reference l oad
but with no bra k i n g . 2 . 3 - Bumber hei ght i n the s i mul at ion means the he i ght of front top edge
of umper (see fi g . 2 ) . These hei ghts are for rather l ow . bumpers wi th the re l at i on to the average hei ght , and esp�c i a l l y so with re l at i on to recent mode l s ; but a s we have made no correcti o ns for the vari ati on i n body pos i ti on due to braki ng (whi ch happens i n most cases pri or to impacti ng ) , the chosen hei ghts s ho u l d be c l ose to the average .
3 - Resu l ts .
3 . 1 - Vel oci ty of centre of gravi ty of head upon i mpact ( see tabl e 2 ) It increases as the bumper i s lowered; I t a l s o i ncreases as bumper projects more .
3 . 2 - Head S . I . ( see tab l e 3 ) The S.I. was ca l cul ated , but the va l ues obta i ned s how that i t i s not a val i d cri teri on for rat i ng bumper hei ghts , for i t depends too much on the s hape and characteri s t i cs of the bodywork where head i mpact takes p l ace for a general concl u s i on to be draw� . On the other hand , the maxi mum vel oci ty of head a l ong i ts trajectory wou l d have been i n teresti ng to have , but i t wasn ' t yet cal cul ated when the s t udy was conducted .
54
h
Angle �onvention for the rotation of the lower limbs and th1� torso of the dununy
normal height for thi s car
normal bumper protruding bumper������� normal hei9ht for this car
Fig . ; 2 ßl e e i ght compared configurations for bumper adj ustement
55
3 . 3 - Maxi mum bumper force . l t varies very l i tt l e and i ncreases s l i ght ly wi th bumper hei g h t . A t a gi ven hei ght , the maximum force i s greater for the square profi l e wi th a non -projecti ng bumper than for the l ow streaml i ned profi l e , and a l so greater for the square profi l e wi t h a projecti ng bumper than for the square profi l e wi th a non-projecti ng bumpe r . The maxi mum force app l i ed by the bumper i s i nteresti ng from the b i o mechan i cal p o i n t of v i ew i nasmuch a s i t i s i n d i rect rat i o w i t h the l ocal i nj ur i es susta i ned . The 2nd . effect of the force appl i ed by the bumper to the l ower members i s to throw them forward and make them turn . The va l ues wh i ch best characteri ze , accord i ng to us , the repercuss i on of the k i nemat i cs are the hori zontal et verti ca l components of the dummy momentum i mparted by the bumper . We i nc l uded th i s cal cul at ion i n the programme after t h i s study , whi ch s howed up the need for i t .
3 . 4 - Rotat i o n of l ower members and torso recorded at 200 ms , consequently after the head-torso/veh i c l e impacts . Genera l l y speaki ng , dummy rotat i o n i n d i recti on of veh i cl e ( see fi g . 3 ) i s greater and qu i cker ( from 1 to 2 rad/s a t 200 ms ) , and the l egs are h i gher i f the bumper i s 40 cm above the ground i ns tead of 50 cm . I s the who l e body turn i ng and the l egs ri s i ng a favourabl e or unfavourab l e factor for the pedestr ian ( effect of head/ground i mpact) ? A conc l u s i on cannot be eas i l y drawn , for , on the one hand , these rotati ons are probably exagerated i n a l l these s imul ati ons ( tooe l as t i c contact components i n the data ) , and on the other han d , the subsequent k i nemat ics of the pedestri a n are l arge ly condi ti oned by the torso/bonnet contacts wh ich occur duri ng the l ong phase precedi ng the fa l l to the ground when the pedestri a n bears more or l ess the bonne t .
3 . 5 - Locat ing the head/vehi cl e i mpact Impact points move very little (from O to 3 cm) and do not a lways go i n the same di rect i o n when bumpe� hei ght i s mod i fi ed .
Howeve r , a s l i ght tendency for t he i mpact poi nt to move to the front of the vehi c le can be detected when the bumper i s l owered . Paradox i c a l l y , a ( s l i gh t ) backward movement of i mpact poi nt i s observed when bumper projection i s i ncreased ( o n l y the square profi l e concerned by thi s ) . Thi s can be expl a i ned by the fact that the greater ang l e formed by the t i gh bones wi th the verti ca l duri ng the t ime they are i n contact wi th front edge of bonnet tends to i nduce their s l i pp i ng backwards .
4 - CONCLUSIONS
4 . 1 - I nfl uence of bumper height on the head/veh i c l e i mpact .
I n both profi l es studied when the bumper does not project , we obtai n , for the head , s l i ght ly better res ul ts w i th a bumper a t 50 cm than at 40 or 35 cm . The di fferences i n i mpact vel oci ty are probably amp l i fi ed by the too great e l asti c i ty of the contact components defi ned i n the data . I n practi ce , they m i ght be compensed for by smooth i ng out the more agress i ve i mpact areas , at the same t ime l oweri ng the bumper .
56
We can ask oursel ves whether the res ul ts obtai ned wi th the model are val i d as far as absorbed energy d i s tr i buti on i s concerned . I t woul d appear that as the chrono l ogi cal order and the di fferent l oad and accel eration l evel s i n the model wi th rel at ion to the test are compl i ed w ith ( wi th the previ ous ly mentioned reservat ions ) , the cal cul a t i on s houl d convey real i ty fa i thful ly .
,., „.
„„
Fi g . 3 I nfl uences of a ) bumper hei ght b ) bumper protrus i on
on pedestri an k i nemati cs .
a )
„ „ .
r-„. b )
57
I t shou ld be noted that l owering the bumper makes the bu l l et car l ess agress i ve i n s i de col l i s i ons .
4 . 2 - I nfl uence of bumper projection on head i mpact severi ty . If the S.I. and max1 �um acceleration of the head are excl uded for the reasons stated i n § 3 . 2 . page 2 , only the head/veh i cl e i mpact vel oci ty remai ns as a pos s i b l e severi ty cri teri on . Accordi ng to thi s cri teri on , no s i gn i fi cant di fference i s observed between both confi gurati ons of bumper i f i t i s at 50 cm ( "norma 1 11 hei ght for the square profi l e used) However , the advantage fal l s cl early to the "non-projecti ng" configurat i on i f the bumper is i n the l ow pos i ti on ( 40 or 35 cm) . Remark I nfluence of bumper projecti on on the severi ty of the femur/l eadi ng edge of bannet i mpact , as wel l as bannet acce l erat i ons . For the three hei ghts i n questi on , the forces generated a t femur/ l eadi ng edge of bannet i mpact , as wel l as the accel erati ons s usta i ned by the pel vi s , are not so h i gh w i t h the projecting bumper than wi th the non-projecti ng bumper . On these two poi nts , the "project ing bumper" confi gurat i on i s therefore better than the "non-project i n g bumper " confi gurat i on .
4 . 3 - General l y speak i ng , there i s no bumper height and projecti on su itab le to a l l veh i c l es . These parameters shou l d be optimi zed as a functi o n of the veh i c l e profi l e and the l ocati on of the area we most wi sh to be i mpacted by the pedestri an ' s hea d .
4 . 4 - The purpose i n l ower i ng the bumper i s to reduce the severi ty o f i njuries to the l ower members whi ch , a l though they are rarely fata l , g i ve ri se to concern by thei rfrequency and fhe d i sabi l i ty they too often g i ve r i se to . A l though at the present t ime i t exi sts no i nj ury cri terion for the l ower members of pedestri an dummi es MVMA/APR , the mathemati cal model wi l l enabl e however knee accel eration as wel l as femur compress i on forces as a functi on of t ime to be obta i ned . Knee acce l erat i on cou l d temporari ly be used . Furthermore , i t wou l d be easy for the knee shearing force to be cal cul ated i nto the model .
4 . 5 - I t wou l d be useful to s i mu l ate other veh i c l e profi l es and vel oci t i es i n order to veri fy whether the cal cul ated val ues deve lop i n the same sense when the height or project i o n of the bumper i s vari ed .
58
TABL
E I
-HE
IGHT
OF
BUMP
ERS-
RESU
LTS
OF T
HE M
ATHE
MATI
CAL
MODE
L
Low
stre
am-
Squa
re p
rofi
l e
Squa
re p
rofi
le
1 i ne
d pr
ofil
e pr
otru
ding
bum
pers
He
ight
50
cm
40 c
m 50
cm
40 c
m 30
cm
50 c
m 40
cm
35 c
m --
Max.
for
ce o
n bu
mper
s (N
) Ma
x. F
x 15
200
1470
0 16
100
1570
0 15
600
1640
0 16
000
1580
0 Ma
x Fz
20
00
3600
21
00
4100
41
00
2300
42
00
4200
Ma
x. r
esul
tant
15
300
1510
0 16
300
1620
0 16
100
1660
0 16
500
1640
0
Wind
shie
ld
Bonn
et o
r sc
uttl
e He
ad i
mpac
t-Ar
ea :
.
Loca
liza
tion
(se
e fi
g.)
24,1
21
,9
1 102,
1 10
1,9
100
,5
1 102
,4
103,
8 10
1,0
. Ti
me
150
145
120
120
120
135
130
130
Velo
city
of
head
C.G
. at
hea
d/
vehi
cle
impa
ct (
m/s)
:
. Vx
8
,29
8,5
2 4
,06
3,7
9 3
,74
3,3
6 4
,73
3,8
8 .
Vz
7,87
8
,60
10,0
8 10
,63
10,9
8 10
,43
11,4
1 11
,87
. Re
sul t
ant
11,4
3 12
, 11
10,8
7 11
,29
11,6
0 10
,96
12,3
5 12
,49
O Max
( g)
:
. He
ad
90
93
121
118
111
117
133
117
. Th
orax
57
58
58
59
61
61
67
67
.
Pelv
is
24
24
65
57
52
48
40
44
Seve
rity
i nd
ex
. He
ad
1071
13
13 1
1419
15
17
1476
1 14
21
1977
17
64
. Th
orax
32
6 36
3 32
6 35
4 41
8 35
0 45
4 49
3
Lowe
r li
mbs-
t =
200
ms
Leg
angl
e (0
) -2
18
-234
-2
15
-229
-2
37
-180
-2
27
-239
Th
igh
angl
e (0
) -2
14
-233
-2
10
-225
-2
32
-197
-2
24
-237
Leg
rota
tion
al v
eloc
ity(
rd/s
) -6
,53
-8,2
0 -1
0,8
-1
2,7
-12,
1 -1
5,7
-1
2,3
-12
,3
. He
ight
/gro
und
-Kn
ees
107
124
116
128
133
109
132
138
Pelv
is
83
91
95
98
100
97
103
103
. Ve
rtic
al k
nee
velo
city
{m/
s}
3,1
5 3
,42
3,8
0 3
,56
3,6
8 2
,32
3 • 7
5 3
,92
t.T1
Rota
tion
al s
peed
of
the
1 Thora
x 5,
52
4,5
7 2
,34
2,5
3 1,
62
4,0
1 2
,08
1,12
l.
.O to
rso
t =
20
0 ms
:
Pelv
is
3,7
5 2
,53
4,2
2 4
,29
3,4
7 6
,27
3,9
8 2
,09
TABLE 2 - VELOC ITY OF THE CENTER OF GRAV ITY OF THE HEAD AT I MPACT
I t i ncreases when the bumper i s l owered : Profi l e Hei ght o f the bumper Ve l oci ty of the
(cm) head ( c . d .g. ) m/s
Low stream l i ned 40/50 1 2 . 1/ 1 1 . 4
Square ( bumper 40/50 1 1 . 3/ 10 . 9 no protrud i n g } 3 5/40 1 1 . 6/ 1 1 . 3
35/50 1 1 . 6/ 10 . 9
Square ( p rotrud i ng 40/50 1 2 . 3/ 1 1 . 0 bumper) 35/40 1 2 . 5/ 12 . 3
35/50 12 . 5/ 1 1 . 0
I t i ncreases too when the bumper i s protrud i ng VELOCITY OF THE CENTER OF GRAVITY OF THE HEAD
Hei ght of the Square profi l e w ith Square profi l e wi th Bumper protrudi ng bumper bumper no protrud i ng
50 cm 1 1 . 0 10 . 9 40 cm 12 . 3 1 1 . 3
35 cm 12 . 5 1 1 . 6
TABLE 3 - S . I . FOR THE HEAD .
l t i ncreases when the bumper i s l owered from 50 to 40 cm .
%
+ 5 . 9
+ 3 . 7 + 2 . 8 + 6 . 7 + 12 . 7 + 1 . 2 + 1 4 . 0
3
+ 0 . 8
+ 9 . 4 + 7 . 7
Profi l e Hei ght of the bumper S . I . Head % ( cm ) ·
Low streaml i ned 40/50 1313/1071 + 22 . 6
Square ( bumper 40/50 1 5 1 5/ 1419 no protrud i ng ) + 6 . 9
Square ( protrud i ng 40/50 1977 /1421 bumper ) + 39 . 1
l t decreases for the two confi gurati ons where 35cm was tested , when the bumper i s s ti l l l owered from 40 to 3 5cm .
Profi l e Hei ght of the bumper S . l . Head %
Square ( bumper no protrud i n g )
Square ( protrudi ng bumper )
( cm )
35/40 35/50
35/40 35/50
1476/ 15 1 7 1476/1419
1 764/ 1977 1 764/ 1421
- 2 . 7 + 4 . 0
- 10 . 8 +24 . 1
60
OaN
8 0 0
6 0 0
4 0 0 2 0 0
5 0
4 0
3 0
2 0
1 0
OaN
1000
600
200
Upper shou lder-Restraint force
'/ I
50 1 0 0 ms
Head and thoracic accelerations
OaN 8 0 0
6 0 0
4 0 0
2 0 0
Lower torso be l t -Re s t r a i n t for�e
5 0
', . .
1 0 0 ms
Dummy run and mathema tical model compared
lit (jl
. Head
50 100
Femur compression loads
r• ' : .
. .r l 50 100 ms
ms
4 0
3 0
2 0
1 0
� ... l9J Thorax
,,..
. I / I
i / / . '
50
Dummy run Model
Fig : 4 , 5 e t 6
100 ms
6 1
A P P E N D I X I I EXAMPLE OF RESULTS FROM MVMA / APR MODE L . Quest ion : Where docs the k i neti c energy go o f a passenger restrai ned by a
bel t and knee stop i n a head-on col l i s i on ? The s i mul ated test i s för a pass en9cr d t 1r:niy i n a P. . 5 . br1dy on a s l ed a t 50 km/hr . The resu l ts of the tcst a n ci li1r1 c!�:1 i'l :" C! r.omparcd ( Fi g .t , : , '7 ) .
- Head and thorax accel erat i ons are s i m i l ar . - The femur l oads are h1 gher i n the tcs t : the knee stop characteri sti cs
are different (Knee bar on r i g i d fi xture i n tes t , model i zed l oad/deformnt ion curve not so s teep ) .
- The l ower shoul der bel t forces are comparabl e . - The upper shoul der be lt forces are hi gher and ha ve a d i fferent shape . I t
i s a problem related to dummy rotati on and sl i ppi ng away at the enrl o f col l i s i on i n the real test : the shape of the upper shoul der bel t force recorded i n the model i s cl ose to the tests vli th no s l i p of tho be l t over the s houl der . The s i mu l a t i on of be l t passage o vcr the thorax ( two predetermi ned poi nt s ) can be quest i oned .
Based on t h i s mode l i zed tes t , the di stri but ion of dummy ki neti c energy duri ng the col l i s i on was ca l cul ated . In the fo l l owi ng tabl es one wi l l f ind :
- The dummy energy whi ch goes i nto structure ( front un i t ) , broken dO\·m d e pendi ng on the forces act i ng on the dummy and wh i ch bri ng about d i s s i pa t i on i n the structure .
- The dummy energy wh ich i s absorbed by the di fferent passenger compartment parts ( be l t + anchor poi nts + scat , knee bar , fl oor) , and by the dummy j o i n ts .
Thi s energy has been d i v i ded i nto two t i me phases of 0 to 70 ms ( end of the body Y , but s l i ght movement of the dummy rema i n s ) , and 70 to 150 ms . The model i zed thorax i n the cal cu l a t i on i s r i g i d .
DUMMY ENERGY ABSORBED BY FRONT U N H
t (ms ) Seat Fl oor Knee stop Be l t TotCI 1
0-70 375 681 1 2 20 698 29 7/!.
70- 1 50 5 10 9 108 1 3 2
0- 150 380 69 1 1 229 eos 3 J.OG
DUMMY ENERGY ABSORBED I N PASSENGER COMPARTMENT Deformations
t (ms ) Seat Fl oor Knee stop Be l t + Total fi xat. i ons
0- 70 247 158 7?.G 15()� ?G:?IJ. 70- 150 52 0 7J 235 3 G G 0-150 299 1 58 805 1 738 3000
62
Fri cti ons and dummy jo i nts
t (ms ) Seat Fl oor Knee stop Be l t Joi nts Total
0-70 164 42 40 40 210 496
70- 150 6 1 26 96 512 641 0- 150 170 43 66 136 722 1137
Ki neti c energy of dummy i s 7222 J at O ms 1 161 J at 70 ms
88 J at 150 ms
whi ch gi ves the fo 1 1 owi ng vari ations 0- 70 ms 606 1 J
70- 150 ms - 1073 J
0- 150 ms - 7134 J
Taki ng account of the reservations made on the qual i ty of s imul ation , the fo l l owing remarks can be made concerni ng the di stri buti on o f dummy energy duri ng the col l i s i o n :
- the 7 kJ for the dummy are spl i t up i nto 3 kJ for the front uni t and 4 kJ i n the i mpact and various fri cti ons i n the passenger cel l . These 4 kJ are di vi ded i nto 3 kJ for deformati on ( i mpacts ) and 1 kJ for fri ction ( dummy/passenger cel l and shoul der be l t + dummy j o i nts ) .
- The knee bar absorbs l ess energy than i t transmi ts to the front uni t ( 871 as aga inst 1229 J ) . I ts l ayout i n the passenger cel l affords good r ide-down . 29% of dummy energy i s transmi tted i n the knee bar forces ( 17% i n front un i t + 12% by knee bar deformati on ) . I t woul d seem, however , that the energy .di ss i pated i n the knee bar i s greater i n the test than i n the s imul ati on . T h i s coul d be due to the characteri s t i cs of the knee bar defi ned as data i n the model ( cf . § 1/ , maximum upper femur l oad on test/mode l ) .
- The 1 1shoul der bel t + anchor poi nt 11 system absorbs more energy than i t transmtts to the front u n i t ( 1874 a s agai nst 806 J ) . Most o f the energy absorbed by the bel t i s duri n g the decel erati on phase of the veh i c l e ( 0 - 7 0 ms ) . I n the case under exami nati on , 30% of the energy coup l ed wi th the force acti ng through the bel t i s transmi tted to the front uni t . 37% o f dummy energy goes through the bel t l oads ( 26% i n the bel t + 1 1% i n front un it ) .
- the seat absorbs n notii1cons i derab l e amount of dummy energy ( 469 J ) and transmi ts another quantity ( 380 J ) to the front uni t . I n a l l , 11% of the energy absorbed goes through the forces app l i ed to the seat , whence the importance of seat fi x ing to fl oo r .
- besi des the energy absorbed by the dummy 1 s joi nts and by be l t fri cti on , most of the energy i s absorbed duri ng the decel erati on phase of the veh i c l e .
63
A P P E N D I X I I I
EXAMPLE OF UTI L I ZAT ION OF THE MVMA/APR MODEL :
Compari son of vari ous so l utions usa b l e for rccons tructi ng real -l i fe acci dents od pedestri ans 1vi th a 50-percenti l e dur.imy to s i rr.ul ate a srnal l er and/or l i ghter vi cti m .
When a car col l i des wi th a pedestrian , the s i ze ( and to a l es ser extent , the wei ght ) of the pedestri an i s of noti ceabl e effect on the head i mpact l ocat i on , and consequent ly on the i nj uri es . I t wou l d therefore appear i ndi s pensabl e when reconstructi ng acci dents u s i ng the 50-percenti l e dummy , to modi fy the s i ze of the dummy (or at l east i ts hei ght wi th rel at i on to the veh i c l e ) to ens ure that i t corrcsponds with that of the v i cti m . To do thi s , several so l utions are pos s i b l e and were s imul ated by the MVMA/APR mode l . The probl em of the wei ght i s con s i dered l ater .
1 - S IMULATED SOLUT IONS . I n an i ni ti al seri es , the authors have stri ven to s i mu l ate an acci dent wh i ch occured between a Peugeot 204 and a fcmal e 10 centi metres l es s i n hei ght than the 5C-percenti l e ma l e , and of about the s ame wei ght .
The s i x so l uti ons mode l l ed are as fol l ows :
1 . 1 - Modi fi ed dummy
Test 1-A : Reduction i n the l engh of the l egs and t i ghs of 5 cm , res ult1 ng i n an overal l reducti on in height of 10 cm . Reducti on i n the i nert i a moments and masses , proporti onal to the reducti ons i n l engh , i n the parts c.oncerned . Test 1-B : Same modi ficati on to the l engh , but w ith the masses and i nert 1 a moments of the l ower l i mbs restored to the i r i n i t i a l val ues , i . e . those of the 50-percenti l e dummy . Tes t 1-C : Shorten i ng of the l c9s and t i ghs by on ly 2 . 5 centimetres , but combi ned with 5 centi metrcs on thc l ower l urnbar of the trunk , consequently amounti ng to 10 cent im�tres i n al l . The ma s ses and i nert i a moments are aga i n kept to thc i ni ti a l val ues .
1 . 2 - Re l ati ve pos i t i on between modi fi ed dummy and vehi c l e . Tes t 2-A : Lower l i mbes ben t , l oweri ng the trun k and head by 10 cent 1metres .
Test 2-B : Dummy wi th l egs stra i ght , 1·1h i l e rai s i ng the vehi cl e ( by changing i ts seat i n g ) by 10 centimetrcs . Tes t 2-C : A compromi se bctwer.n tcst 2-A and test 2-B : s l i ght bend-1 ng of the l egs ( he i ght of hcad rcduced by 5 centi metres on ly ) , whi l e a t the same t ime rai s i n g the vch i cl c by 5 cent imetres .
2 - I NPUT DATA TO MATHEMAT I CAL MODEL .
The vehi cl e profi l e used was that of a Peugeot 204 a l ong the centre - l i ne of the headl amps , j u s t as i n the real - l i fe acci dent .
64
I n th i s series of s i mu l ati ons , a parti cu lar ri gi d i ty was not gi ven to the scutt l e area . Thus , the ri gi d i ty of the bonnet was used up to the wi ndscreen . The head accel erati ons , the severi ty i ndex ( S I ) and the ( H I C ) depend therefore only o n the speed of the head at i mpact aga i n st the vehi cl e . The att itude of the veh i c l e was that of the unbraked vehi c l ed carrying a standard reference l o ad . The reason for thi s i s that duri ng the reconstructi on , i t was des i gned to brake the veh i c l e only a t fi rst impact , i n order better to control the hei ght of the nose of the bonnet on i mpact . Two other cases were cons i dered i n order to act as references , namely :
- Test 0 , w ith the 50-percenti l e dummy i n the normal pos i ti o n . - Test Oa , with an homothet i c dummy reduced a s to be 1 0 cm s horter , and with
the same mas s .
3 - RESULTS ( see tabl e4appended) I Al l the tests wi th the trunk at i ts i ni ti al l engh ( i . e . al l tests except 1-C) res u l t in an impact poi nt from 83 to 87 . 5 centi metres from the nose of the bonnet , as agai nst 1 10 wi th the normal 50-percenti l e dummy ( test 0 ) , i . e . a mean s h i ft of about 25 centimetres . Tes t 1-C s houl d be compared to test Oa , s i n ce i n both cases the trunk was shortened , and i n order to have the same head height , the pe l vi s occupied a hi gher pos i tion wi th rel at ion to the veh i c l e . S i n ce the edge of the bonnet stri kes the dummy l ower compared to i ts centre of gravi ty and at the h i p of the dummy , the trunk rotates more s l owly , the dummy moves further towards the rear . The speed wi th whi ch the head s tri kes the bonnet i s l ower and the poi nt of impact further bac k . ( The reader may be remi nded that the scuttl e at thi s poi n t has the same r i g i d i ty as the bonnet , so o ne cannot concl ude that the impact i s any l es s severe i n cases 1-C or Oa , than i n the other cases . ) Apart from case 1 - C , the resu l ts are fai rly s i mi l ar . Confi guration 2-A ( dummy wi th l egs bent ) bei ng �he eas i est to carry out u s i ng concerned test faci l i ti e s , seems to be preferab l e , when the 50-percenti l e mal e dummy must be used to reconstruct a rea l - l i fe pedestrian acci den t , the vi ctim of wh i ch i s s l i ghtly s horter i n hei ght for the same wei ght .
4 - CONFI RMATION BY MEANS OF CATAPULT TESTS A test with a dummy was made in confi gurat i on 2-A . The resu l ts are very s imi l ar to those obtai ned by the mathematical model . I n parti cul a r , the point of i mpact i s the same to wi thi n 1 centimetre ( see tabl e of resu l ts ) . A second test was made , wi th the dummy rai sed by 4 centimetres ( s ti l l with the l egs bent) , i n order to approximate more c l ose ly to the actual cond i t i ons of the acci dent , i n whi ch the driver stated that he had braked before impact , whi ch wou l d thus d i p the nose of the vehi cl e . I n thi s cas e , the head s truck the scutt l e i nstead of the true acci dent . Thi s test was s imu l ated i n the case 2-A-b , where a scuttl e repl aced the secti on of bonnet at the bottom of the wi ndscreen . Here agai n , the res ul ts are h i gh ly s i mi l ar .
65
"'
"'
iest
n'
0 Oa la
lb
k
2�
Q
lt
�
E-<
2c
:::>
e:::
2Aa
�
0
2Ab
u
lDa
l
lDb
lDc
lOd
--
-2
a,
dull'fl
ly
2Ab,
du
rrmy
TA8
LE
4 -
COMP
ARED
RES
ULTS
DF
COM
PUTE
R AN
D DU
MMY
RUNS
CONF
1 GU
RA T
I ONS
HE
AD
PEA
K s.
I.
H. I.
C.
g' s
ti
me
{ms)
--
50%
dull
llly
115
12
5 15
72
1228
Homo
thet
ic d
ummy
1.65
ta
ll
71
120
10
23
804
Shor
ter
legs
-lo
wer
we
ight
s 12
2 10
2.5
14
45
1042
Shor
ter
legs
-
50%
we
ight
s 12
5 10
2.5
14
96
1055
Lum
bar
spin
e an
d le
gs
redu
ced
77
117
.5
1028
8
03
Flex
ed l
egs
119
10
0 14
33
1057
Ral
sed
ve
hicl
e
123
97 . 5
15
32
1133
lnte
rmed
late
bet
wee
n 2a
an
d 2
b
121
')7
.5
1476
10
76
Fle
xed
le
gs,
z
-5
cm
onl
y
99
122
.5
1235
99
2
Flex
ed
legs
-
6 cm
; r
ea
l1st
ic s
cutt
le
152
120
23
17
1817
Like
lc
+ lo
wer
we
tght
s 12
5 11
7 .5
15
08
11
35
Lik
e l
Da
; on
e st
de
wit
hou
t lo
wer
arm
12
6 11
7 .5
15
42
1172
Ltk
e l
Da;
one
sid
e w
itho
ut u
pper
11m
bs
126
117
.5
1477
11
12
Lik
e l
Da
; bo
th s
lde
s w
ith
out
up
per
ltm
bs
123
117
.5
1449
10
65
2a p
aram
eter
s -
test
wit
h du
1m1y
12
2 11
0
--
2Ab
para
mete
rs
-te
st w
lth
dur
rmy
il!
6 11
3 24
24
1615
1 O
z
dis
tanc
e fr
om h
ead
impa
ct p
oin
t to
the
fr
ont
edge
of
the
bonn
et.
11
Tho
raci
c ar
e to
o h
igh
acco
rdin
g to
mat
hem
ati
cal
mode
l.
Che
st s
tiff
ness
ha
s to
be
redu
ced
in mo
deli
ng
=•
2 C
ar s
peed
: 11
.11
m/s
(4
0 k
m/h
)
D cm
{ lt
)
110
96
87 .
5
87
95
87
83
84
104
.5
102.
6
95
95 93
.7
91
. 5
86
98
THOR
AX
PEA
K H
IP
PEA
K HE
AO/
CAR
SPEE
D
g's
{:;,
1t)
time
g
's
tim
e a
t i
mpa
ct m
/s
{1t1t1t
) --
--56
13
0
94
15
11.6
6
53
120
10
7 10
10
.99
59
105
112
10
12
.25
60
105
117
10
12
.19
51
115
10
2 15
1
1.3
5
54
105
113
10
12
.41
58
100
110
15
13
.04
57
100
113
10
12
.95
57
125
104
15
11.8
3
79
120
.1
05
15
11
.66
67
.5
115
10
0 15
10
.94
66
115
100
15
11.2
5
68
115
101
15
10.6
5
68
115
10
4 15
10
.63
27
125
102
18
30.5
12
0 8
3 13
5 - OTHER CONFIGURATI ONS S IMULATED . A seri es of four s imul at ions was ccnducted , �eri ved from the 1-C case , but wi th the masses and moments of i nert i a of the l ower l i mbs and l ower thorax modi fi ed i n proporti on to the l ength . r � addi t i on , a l l or part of the upper members were wi thdrawn i n these four tests , these modi fi cat i ons bei ng easy to carry out on a reference dummy . The purpose was to approximate the 50�� ma 1 e dummy to a subject s horter by 10 cm and s l i ght ly l i ghter , wi th fai rly harmoni ous di stri bution of the wei g hts .
These four s imul at ions are 1-D-a Same d imensi ons and confi gurat i ons than i n 1-C but w i th the masses
and moments of i nert i a of the l egs and l ower thorax reduced i n proport i on to thei r l en gths . Tota l wei ght : 72 . 1 kg i nstead of 74 . 9 kg for the 50% dummy .
1-D-b 1-D-c
As for 1-D- a , w i th one forearm el i mi n ated . Total wei ght : 69 . 7 kg . As for 1-D-a , w ith one upper l i mb el � m i nated . Total wei ght : 67 . l kg .
--
1-D-d As for 1-0- ? , wi th both upper l i mbs el i mi n ated . Total wei ght : 62 . 2 kg .
The ri g i di ty of the scutt l e i s a ga i n equal to that of the banne t . The resul ts approximate to those o f test 1 - C a s regards the pos i t i on o f the head-to-bonnet i mpact poi nt and the H I C .
I t i s merely to be observed that the head-to-bonnet i mpact poi nt moves s l i ghtly toward the front of the veh i cl e , as the upper l i mbs a re el i mi nated. Th i s can be expl a i ned by the fact that the l ower i nert i a o f the upper part o f the dummy cuts down the s h i ft towards the rea r .
6 - CONCLUSI ONS 6 . 1 - Real - l i fe acci dents to pedestri ans the he ight of whi ch i s shorter , tliOugh fa i rl y c l osed to that of the 50% ma J e dummy , can be constructed sat i s factori ly by such a dummy , provi ded the hei ght of the top of the head be brought to coi nci de w ith the hei ght of the pedestr i an i nvo l ved , s imply by bendi ng i ts l egs .
6 . 2 - I f only the head i mpact l ocation i s des i red , i t wou l d not appear necessary to adjust the wei ght to that of the v i ct i m , s i nce t h i s wei ght has but l i tt l e effect on the trajectory . Th i s concl us ion does not apply i f comparab le va l ues of the H I C are des i red . 6 . 3 - Thi s app l i es at l east to pedestri ans of normal corpul ence and for �eh i c l e profi l e comparable to that of the Peugeot 204 . A s im i l ar study wi l l be necessary to confi rm these resu l ts for di fferent profi l es . 6 . 4 . - The pract ica l l i mi t to the method , i . e . the maxi mum d i stance the dummy can be l owered wi thout i ntroducung unacceptabl e errors , must a l so be d i scovered .
67